Azure DevOps Connector via Model Context Protocol (MCP)

This repository offers an implementation of an MCP server designed to serve as an intelligent intermediary, allowing large language models (LLMs) to interface directly with Azure DevOps functionalities through conversational commands.

System Overview

This tool bridges the gap between human language instructions and the underlying Azure DevOps REST API endpoints. It translates abstract user requests into concrete, executable API calls, supporting comprehensive lifecycle management within the development ecosystem.

Core Capabilities

Artifact Lifecycle Management (Work Items)

Retrieval & Scrutiny: Execute complex Work Item Query Language (WIQL) statements to fetch item sets. Detail examination of any specific item's attributes.
Creation & Ingestion: Provision new artifacts (e.g., Tasks, Bugs, Epics) populated with initial data.
Modification & Revision: Update extant item fields, status transitions, and attach supporting narrative via comments.
Hierarchical Structuring: Define and manipulate parent-child linkages between related work items.

Organizational Structure Interaction

Project Inventory: Enumerate all accessible Azure DevOps projects.
Team Visibility: List organizational teams and query their membership rosters.
Temporal Context: Access configurations for team sprints and iteration paths.
Scope Definition: Determine the Area Path assignments associated with specific teams.

Forthcoming Enhancements (Roadmap)

CI/CD Orchestration: Capabilities to query build/release pipeline states and initiate new execution sequences.
Source Control Collaboration: Support for managing Pull Request creation, review assignments, and state modifications.
Agile Cadence Control: Tools for sprint planning and backlog grooming operations.
Repository Governance: Administration features for setting mandatory branch protection rules.

Deployment and Setup

Prerequisites for Operation

A functional environment running Python version 3.10 or newer.
Active credentials for an Azure DevOps tenant.
A generated Personal Access Token (PAT) configured with the necessary scope permissions for the target APIs.

Installation Procedure

bash

Obtain the source code

git clone [repository_url] cd mcp-azure-devops

Install dependencies in editable mode (for development)

uv pip install -e ".[dev]"

Standard installation via package registry

pip install mcp-azure-devops

Configuration Directives

Environment variables must be defined, typically within a .env file in the root directory, to establish secure connectivity:

AZURE_DEVOPS_PAT=your_secure_token_here AZURE_DEVOPS_ORGANIZATION_URL=https://dev.azure.com/YourOrganizationName

Ensure the organization URL is complete.

Server Execution

To initiate the MCP server for AI integration:

bash

Launch for local testing with the MCP diagnostic utility

mcp dev src/mcp_azure_devops/server.py

Deploy for use with client applications (e.g., Claude Desktop)

mcp install src/mcp_azure_devops/server.py --name "ADO Agent Connector"

Usage Scenarios (Illustrative Prompts)

Querying Work Items

"Retrieve all outstanding defects assigned directly to me that are targeted for the current iteration cycle."

Creating an Artifact

"Provision a new Epic within the 'Alpha-Project' repository, titled 'System Security Overhaul', slated for Q4 delivery."

Updating and Annotating

"Mark work item ID 4501 as 'Done' and append a note confirming successful validation of the associated code changes."

Team Structure Inquiry

"What are the designated members of the 'Platform Services' group residing under the 'Project Delta' structure? Also, list their primary area assignments."

Internal Architecture Notes

The codebase is segmented by functional domain for modularity:

features/work_items: Logic governing item CRUD operations and querying.
features/projects: Methods dealing with organization, team, and structural elements.
features/teams: Specific utilities for group membership and scheduling retrieval.
utils: Shared components, primarily for API client instantiation and connection handling.

Detailed protocol specifications are located in CLAUDE.md.